Abstract:

Text-to-3D generation is an emerging field that aims to synthesize three-dimensional (3D) objects, scenes, and environments from text descriptions. This task leverages the progress in deep learning, notably in generative models, neural rendering, and large-scale vision-language pretraining. 3D content generation from text descriptions has significant applications in gaming, virtual reality, product design, and digital content creation. This survey provides a more extended overview of recent breakthroughs in text-to-3D generation and discusses the ability of textto- 3d models to generate scenes. The review attempts to provide scientists and practitioners with a structured understanding of current trends and potential future breakthroughs in this field. One of the most exciting applications of this potential is in the area of 3D printing, allowing levels of design creativity and optimization heretofore unknown. By leveraging these techniques, designers are able to push the limits of innovation, both aesthetically and functionally.